mbed official / mbed-dsp

CMSIS DSP library

Dependents:   performance_timer Surfboard_ gps2rtty Capstone ... more

Legacy Warning

This is an mbed 2 library. To learn more about mbed OS 5, visit the docs.

cmsis_dsp/FilteringFunctions/arm_conv_partial_opt_q15.c@1:fdd22bb7aa52, 2012-11-28 (annotated)

Committer:
emilmont
Date:
Wed Nov 28 12:30:09 2012 +0000
Revision:
1:fdd22bb7aa52
Child:
2:da51fb522205
DSP library code

Who changed what in which revision?

UserRevisionLine numberNew contents of line
emilmont 1:fdd22bb7aa52 1 /* ----------------------------------------------------------------------
emilmont 1:fdd22bb7aa52 2 * Copyright (C) 2010 ARM Limited. All rights reserved.
emilmont 1:fdd22bb7aa52 3 *
emilmont 1:fdd22bb7aa52 4 * $Date: 15. February 2012
emilmont 1:fdd22bb7aa52 5 * $Revision: V1.1.0
emilmont 1:fdd22bb7aa52 6 *
emilmont 1:fdd22bb7aa52 7 * Project: CMSIS DSP Library
emilmont 1:fdd22bb7aa52 8 * Title: arm_conv_partial_opt_q15.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 1:fdd22bb7aa52 10 * Description: Partial convolution of Q15 sequences.
emilmont 1:fdd22bb7aa52 11 *
emilmont 1:fdd22bb7aa52 12 * Target Processor: Cortex-M4/Cortex-M3
emilmont 1:fdd22bb7aa52 13 *
emilmont 1:fdd22bb7aa52 14 * Version 1.1.0 2012/02/15
emilmont 1:fdd22bb7aa52 15 * Updated with more optimizations, bug fixes and minor API changes.
emilmont 1:fdd22bb7aa52 16 *
emilmont 1:fdd22bb7aa52 17 * Version 1.0.11 2011/10/18
emilmont 1:fdd22bb7aa52 18 * Bug Fix in conv, correlation, partial convolution.
emilmont 1:fdd22bb7aa52 19 *
emilmont 1:fdd22bb7aa52 20 * Version 1.0.10 2011/7/15
emilmont 1:fdd22bb7aa52 21 * Big Endian support added and Merged M0 and M3/M4 Source code.
emilmont 1:fdd22bb7aa52 22 *
emilmont 1:fdd22bb7aa52 23 * Version 1.0.3 2010/11/29
emilmont 1:fdd22bb7aa52 24 * Re-organized the CMSIS folders and updated documentation.
emilmont 1:fdd22bb7aa52 25 *
emilmont 1:fdd22bb7aa52 26 * Version 1.0.2 2010/11/11
emilmont 1:fdd22bb7aa52 27 * Documentation updated.
emilmont 1:fdd22bb7aa52 28 *
emilmont 1:fdd22bb7aa52 29 * Version 1.0.1 2010/10/05
emilmont 1:fdd22bb7aa52 30 * Production release and review comments incorporated.
emilmont 1:fdd22bb7aa52 31 *
emilmont 1:fdd22bb7aa52 32 * Version 1.0.0 2010/09/20
emilmont 1:fdd22bb7aa52 33 * Production release and review comments incorporated
emilmont 1:fdd22bb7aa52 34 *
emilmont 1:fdd22bb7aa52 35 * Version 0.0.7 2010/06/10
emilmont 1:fdd22bb7aa52 36 * Misra-C changes done
emilmont 1:fdd22bb7aa52 37 *
emilmont 1:fdd22bb7aa52 38 * -------------------------------------------------------------------- */
emilmont 1:fdd22bb7aa52 39
emilmont 1:fdd22bb7aa52 40 #include "arm_math.h"
emilmont 1:fdd22bb7aa52 41
emilmont 1:fdd22bb7aa52 42 /**
emilmont 1:fdd22bb7aa52 43 * @ingroup groupFilters
emilmont 1:fdd22bb7aa52 44 */
emilmont 1:fdd22bb7aa52 45
emilmont 1:fdd22bb7aa52 46 /**
emilmont 1:fdd22bb7aa52 47 * @addtogroup PartialConv
emilmont 1:fdd22bb7aa52 48 * @{
emilmont 1:fdd22bb7aa52 49 */
emilmont 1:fdd22bb7aa52 50
emilmont 1:fdd22bb7aa52 51 /**
emilmont 1:fdd22bb7aa52 52 * @brief Partial convolution of Q15 sequences.
emilmont 1:fdd22bb7aa52 53 * @param[in] *pSrcA points to the first input sequence.
emilmont 1:fdd22bb7aa52 54 * @param[in] srcALen length of the first input sequence.
emilmont 1:fdd22bb7aa52 55 * @param[in] *pSrcB points to the second input sequence.
emilmont 1:fdd22bb7aa52 56 * @param[in] srcBLen length of the second input sequence.
emilmont 1:fdd22bb7aa52 57 * @param[out] *pDst points to the location where the output result is written.
emilmont 1:fdd22bb7aa52 58 * @param[in] firstIndex is the first output sample to start with.
emilmont 1:fdd22bb7aa52 59 * @param[in] numPoints is the number of output points to be computed.
emilmont 1:fdd22bb7aa52 60 * @param[in] *pScratch1 points to scratch buffer of size max(srcALen, srcBLen) + 2*min(srcALen, srcBLen) - 2.
emilmont 1:fdd22bb7aa52 61 * @param[in] *pScratch2 points to scratch buffer of size min(srcALen, srcBLen).
emilmont 1:fdd22bb7aa52 62 * @return Returns either ARM_MATH_SUCCESS if the function completed correctly or ARM_MATH_ARGUMENT_ERROR if the requested subset is not in the range [0 srcALen+srcBLen-2].
emilmont 1:fdd22bb7aa52 63 *
emilmont 1:fdd22bb7aa52 64 * \par Restrictions
emilmont 1:fdd22bb7aa52 65 * If the silicon does not support unaligned memory access enable the macro UNALIGNED_SUPPORT_DISABLE
emilmont 1:fdd22bb7aa52 66 * In this case input, output, state buffers should be aligned by 32-bit
emilmont 1:fdd22bb7aa52 67 *
emilmont 1:fdd22bb7aa52 68 * Refer to <code>arm_conv_partial_fast_q15()</code> for a faster but less precise version of this function for Cortex-M3 and Cortex-M4.
emilmont 1:fdd22bb7aa52 69 *
emilmont 1:fdd22bb7aa52 70 *
emilmont 1:fdd22bb7aa52 71 */
emilmont 1:fdd22bb7aa52 72
emilmont 1:fdd22bb7aa52 73 #ifndef UNALIGNED_SUPPORT_DISABLE
emilmont 1:fdd22bb7aa52 74
emilmont 1:fdd22bb7aa52 75 arm_status arm_conv_partial_opt_q15(
emilmont 1:fdd22bb7aa52 76 q15_t * pSrcA,
emilmont 1:fdd22bb7aa52 77 uint32_t srcALen,
emilmont 1:fdd22bb7aa52 78 q15_t * pSrcB,
emilmont 1:fdd22bb7aa52 79 uint32_t srcBLen,
emilmont 1:fdd22bb7aa52 80 q15_t * pDst,
emilmont 1:fdd22bb7aa52 81 uint32_t firstIndex,
emilmont 1:fdd22bb7aa52 82 uint32_t numPoints,
emilmont 1:fdd22bb7aa52 83 q15_t * pScratch1,
emilmont 1:fdd22bb7aa52 84 q15_t * pScratch2)
emilmont 1:fdd22bb7aa52 85 {
emilmont 1:fdd22bb7aa52 86
emilmont 1:fdd22bb7aa52 87 q15_t *pOut = pDst; /* output pointer */
emilmont 1:fdd22bb7aa52 88 q15_t *pScr1 = pScratch1; /* Temporary pointer for scratch1 */
emilmont 1:fdd22bb7aa52 89 q15_t *pScr2 = pScratch2; /* Temporary pointer for scratch1 */
emilmont 1:fdd22bb7aa52 90 q63_t acc0, acc1, acc2, acc3; /* Accumulator */
emilmont 1:fdd22bb7aa52 91 q31_t x1, x2, x3; /* Temporary variables to hold state and coefficient values */
emilmont 1:fdd22bb7aa52 92 q31_t y1, y2; /* State variables */
emilmont 1:fdd22bb7aa52 93 q15_t *pIn1; /* inputA pointer */
emilmont 1:fdd22bb7aa52 94 q15_t *pIn2; /* inputB pointer */
emilmont 1:fdd22bb7aa52 95 q15_t *px; /* Intermediate inputA pointer */
emilmont 1:fdd22bb7aa52 96 q15_t *py; /* Intermediate inputB pointer */
emilmont 1:fdd22bb7aa52 97 uint32_t j, k, blkCnt; /* loop counter */
emilmont 1:fdd22bb7aa52 98 arm_status status; /* Status variable */
emilmont 1:fdd22bb7aa52 99 uint32_t tapCnt; /* loop count */
emilmont 1:fdd22bb7aa52 100
emilmont 1:fdd22bb7aa52 101 /* Check for range of output samples to be calculated */
emilmont 1:fdd22bb7aa52 102 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emilmont 1:fdd22bb7aa52 103 {
emilmont 1:fdd22bb7aa52 104 /* Set status as ARM_MATH_ARGUMENT_ERROR */
emilmont 1:fdd22bb7aa52 105 status = ARM_MATH_ARGUMENT_ERROR;
emilmont 1:fdd22bb7aa52 106 }
emilmont 1:fdd22bb7aa52 107 else
emilmont 1:fdd22bb7aa52 108 {
emilmont 1:fdd22bb7aa52 109
emilmont 1:fdd22bb7aa52 110 /* The algorithm implementation is based on the lengths of the inputs. */
emilmont 1:fdd22bb7aa52 111 /* srcB is always made to slide across srcA. */
emilmont 1:fdd22bb7aa52 112 /* So srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 113 if(srcALen >= srcBLen)
emilmont 1:fdd22bb7aa52 114 {
emilmont 1:fdd22bb7aa52 115 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 116 pIn1 = pSrcA;
emilmont 1:fdd22bb7aa52 117
emilmont 1:fdd22bb7aa52 118 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 119 pIn2 = pSrcB;
emilmont 1:fdd22bb7aa52 120 }
emilmont 1:fdd22bb7aa52 121 else
emilmont 1:fdd22bb7aa52 122 {
emilmont 1:fdd22bb7aa52 123 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 124 pIn1 = pSrcB;
emilmont 1:fdd22bb7aa52 125
emilmont 1:fdd22bb7aa52 126 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 127 pIn2 = pSrcA;
emilmont 1:fdd22bb7aa52 128
emilmont 1:fdd22bb7aa52 129 /* srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 130 j = srcBLen;
emilmont 1:fdd22bb7aa52 131 srcBLen = srcALen;
emilmont 1:fdd22bb7aa52 132 srcALen = j;
emilmont 1:fdd22bb7aa52 133 }
emilmont 1:fdd22bb7aa52 134
emilmont 1:fdd22bb7aa52 135 /* Temporary pointer for scratch2 */
emilmont 1:fdd22bb7aa52 136 py = pScratch2;
emilmont 1:fdd22bb7aa52 137
emilmont 1:fdd22bb7aa52 138 /* pointer to take end of scratch2 buffer */
emilmont 1:fdd22bb7aa52 139 pScr2 = pScratch2 + srcBLen - 1;
emilmont 1:fdd22bb7aa52 140
emilmont 1:fdd22bb7aa52 141 /* points to smaller length sequence */
emilmont 1:fdd22bb7aa52 142 px = pIn2;
emilmont 1:fdd22bb7aa52 143
emilmont 1:fdd22bb7aa52 144 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 145 k = srcBLen >> 2u;
emilmont 1:fdd22bb7aa52 146
emilmont 1:fdd22bb7aa52 147 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 148 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 149 while(k > 0u)
emilmont 1:fdd22bb7aa52 150 {
emilmont 1:fdd22bb7aa52 151 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 152 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 153 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 154 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 155 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 156
emilmont 1:fdd22bb7aa52 157 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 158 k--;
emilmont 1:fdd22bb7aa52 159 }
emilmont 1:fdd22bb7aa52 160
emilmont 1:fdd22bb7aa52 161 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 162 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 163 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 164
emilmont 1:fdd22bb7aa52 165 while(k > 0u)
emilmont 1:fdd22bb7aa52 166 {
emilmont 1:fdd22bb7aa52 167 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 168 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 169
emilmont 1:fdd22bb7aa52 170 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 171 k--;
emilmont 1:fdd22bb7aa52 172 }
emilmont 1:fdd22bb7aa52 173
emilmont 1:fdd22bb7aa52 174 /* Initialze temporary scratch pointer */
emilmont 1:fdd22bb7aa52 175 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 176
emilmont 1:fdd22bb7aa52 177 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emilmont 1:fdd22bb7aa52 178 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 179
emilmont 1:fdd22bb7aa52 180 /* Update temporary scratch pointer */
emilmont 1:fdd22bb7aa52 181 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 182
emilmont 1:fdd22bb7aa52 183 /* Copy bigger length sequence(srcALen) samples in scratch1 buffer */
emilmont 1:fdd22bb7aa52 184
emilmont 1:fdd22bb7aa52 185 /* Copy (srcALen) samples in scratch buffer */
emilmont 1:fdd22bb7aa52 186 arm_copy_q15(pIn1, pScr1, srcALen);
emilmont 1:fdd22bb7aa52 187
emilmont 1:fdd22bb7aa52 188 /* Update pointers */
emilmont 1:fdd22bb7aa52 189 pScr1 += srcALen;
emilmont 1:fdd22bb7aa52 190
emilmont 1:fdd22bb7aa52 191 /* Fill (srcBLen - 1u) zeros at end of scratch buffer */
emilmont 1:fdd22bb7aa52 192 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 193
emilmont 1:fdd22bb7aa52 194 /* Update pointer */
emilmont 1:fdd22bb7aa52 195 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 196
emilmont 1:fdd22bb7aa52 197 /* Initialization of pIn2 pointer */
emilmont 1:fdd22bb7aa52 198 pIn2 = py;
emilmont 1:fdd22bb7aa52 199
emilmont 1:fdd22bb7aa52 200 pScratch1 += firstIndex;
emilmont 1:fdd22bb7aa52 201
emilmont 1:fdd22bb7aa52 202 pOut = pDst + firstIndex;
emilmont 1:fdd22bb7aa52 203
emilmont 1:fdd22bb7aa52 204 /* Actual convolution process starts here */
emilmont 1:fdd22bb7aa52 205 blkCnt = (numPoints) >> 2;
emilmont 1:fdd22bb7aa52 206
emilmont 1:fdd22bb7aa52 207 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 208 {
emilmont 1:fdd22bb7aa52 209 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 210 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 211
emilmont 1:fdd22bb7aa52 212 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 213 acc0 = 0;
emilmont 1:fdd22bb7aa52 214 acc1 = 0;
emilmont 1:fdd22bb7aa52 215 acc2 = 0;
emilmont 1:fdd22bb7aa52 216 acc3 = 0;
emilmont 1:fdd22bb7aa52 217
emilmont 1:fdd22bb7aa52 218 /* Read two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 219 x1 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 220
emilmont 1:fdd22bb7aa52 221 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 222 x2 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 223
emilmont 1:fdd22bb7aa52 224 tapCnt = (srcBLen) >> 2u;
emilmont 1:fdd22bb7aa52 225
emilmont 1:fdd22bb7aa52 226 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 227 {
emilmont 1:fdd22bb7aa52 228
emilmont 1:fdd22bb7aa52 229 /* Read four samples from smaller buffer */
emilmont 1:fdd22bb7aa52 230 y1 = _SIMD32_OFFSET(pIn2);
emilmont 1:fdd22bb7aa52 231 y2 = _SIMD32_OFFSET(pIn2 + 2u);
emilmont 1:fdd22bb7aa52 232
emilmont 1:fdd22bb7aa52 233 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 234 acc0 = __SMLALD(x1, y1, acc0);
emilmont 1:fdd22bb7aa52 235 acc2 = __SMLALD(x2, y1, acc2);
emilmont 1:fdd22bb7aa52 236
emilmont 1:fdd22bb7aa52 237 /* pack input data */
emilmont 1:fdd22bb7aa52 238 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 239 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 240 #else
emilmont 1:fdd22bb7aa52 241 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 242 #endif
emilmont 1:fdd22bb7aa52 243
emilmont 1:fdd22bb7aa52 244 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 245 acc1 = __SMLALDX(x3, y1, acc1);
emilmont 1:fdd22bb7aa52 246
emilmont 1:fdd22bb7aa52 247 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 248 x1 = _SIMD32_OFFSET(pScr1);
emilmont 1:fdd22bb7aa52 249
emilmont 1:fdd22bb7aa52 250 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 251 acc0 = __SMLALD(x2, y2, acc0);
emilmont 1:fdd22bb7aa52 252 acc2 = __SMLALD(x1, y2, acc2);
emilmont 1:fdd22bb7aa52 253
emilmont 1:fdd22bb7aa52 254 /* pack input data */
emilmont 1:fdd22bb7aa52 255 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 256 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 257 #else
emilmont 1:fdd22bb7aa52 258 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 259 #endif
emilmont 1:fdd22bb7aa52 260
emilmont 1:fdd22bb7aa52 261 acc3 = __SMLALDX(x3, y1, acc3);
emilmont 1:fdd22bb7aa52 262 acc1 = __SMLALDX(x3, y2, acc1);
emilmont 1:fdd22bb7aa52 263
emilmont 1:fdd22bb7aa52 264 x2 = _SIMD32_OFFSET(pScr1 + 2u);
emilmont 1:fdd22bb7aa52 265
emilmont 1:fdd22bb7aa52 266 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 267 x3 = __PKHBT(x2, x1, 0);
emilmont 1:fdd22bb7aa52 268 #else
emilmont 1:fdd22bb7aa52 269 x3 = __PKHBT(x1, x2, 0);
emilmont 1:fdd22bb7aa52 270 #endif
emilmont 1:fdd22bb7aa52 271
emilmont 1:fdd22bb7aa52 272 acc3 = __SMLALDX(x3, y2, acc3);
emilmont 1:fdd22bb7aa52 273
emilmont 1:fdd22bb7aa52 274 /* update scratch pointers */
emilmont 1:fdd22bb7aa52 275 pIn2 += 4u;
emilmont 1:fdd22bb7aa52 276 pScr1 += 4u;
emilmont 1:fdd22bb7aa52 277
emilmont 1:fdd22bb7aa52 278
emilmont 1:fdd22bb7aa52 279 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 280 tapCnt--;
emilmont 1:fdd22bb7aa52 281 }
emilmont 1:fdd22bb7aa52 282
emilmont 1:fdd22bb7aa52 283 /* Update scratch pointer for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 284 pScr1 -= 4u;
emilmont 1:fdd22bb7aa52 285
emilmont 1:fdd22bb7aa52 286 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 287 tapCnt = (srcBLen) & 3u;
emilmont 1:fdd22bb7aa52 288
emilmont 1:fdd22bb7aa52 289 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 290 {
emilmont 1:fdd22bb7aa52 291 /* accumlate the results */
emilmont 1:fdd22bb7aa52 292 acc0 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 293 acc1 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 294 acc2 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 295 acc3 += (*pScr1++ * *pIn2++);
emilmont 1:fdd22bb7aa52 296
emilmont 1:fdd22bb7aa52 297 pScr1 -= 3u;
emilmont 1:fdd22bb7aa52 298
emilmont 1:fdd22bb7aa52 299 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 300 tapCnt--;
emilmont 1:fdd22bb7aa52 301 }
emilmont 1:fdd22bb7aa52 302
emilmont 1:fdd22bb7aa52 303 blkCnt--;
emilmont 1:fdd22bb7aa52 304
emilmont 1:fdd22bb7aa52 305
emilmont 1:fdd22bb7aa52 306 /* Store the results in the accumulators in the destination buffer. */
emilmont 1:fdd22bb7aa52 307
emilmont 1:fdd22bb7aa52 308 #ifndef ARM_MATH_BIG_ENDIAN
emilmont 1:fdd22bb7aa52 309
emilmont 1:fdd22bb7aa52 310 *__SIMD32(pOut)++ =
emilmont 1:fdd22bb7aa52 311 __PKHBT(__SSAT((acc0 >> 15), 16), __SSAT((acc1 >> 15), 16), 16);
emilmont 1:fdd22bb7aa52 312 *__SIMD32(pOut)++ =
emilmont 1:fdd22bb7aa52 313 __PKHBT(__SSAT((acc2 >> 15), 16), __SSAT((acc3 >> 15), 16), 16);
emilmont 1:fdd22bb7aa52 314
emilmont 1:fdd22bb7aa52 315 #else
emilmont 1:fdd22bb7aa52 316
emilmont 1:fdd22bb7aa52 317 *__SIMD32(pOut)++ =
emilmont 1:fdd22bb7aa52 318 __PKHBT(__SSAT((acc1 >> 15), 16), __SSAT((acc0 >> 15), 16), 16);
emilmont 1:fdd22bb7aa52 319 *__SIMD32(pOut)++ =
emilmont 1:fdd22bb7aa52 320 __PKHBT(__SSAT((acc3 >> 15), 16), __SSAT((acc2 >> 15), 16), 16);
emilmont 1:fdd22bb7aa52 321
emilmont 1:fdd22bb7aa52 322 #endif /* #ifndef ARM_MATH_BIG_ENDIAN */
emilmont 1:fdd22bb7aa52 323
emilmont 1:fdd22bb7aa52 324 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 325 pIn2 = py;
emilmont 1:fdd22bb7aa52 326
emilmont 1:fdd22bb7aa52 327 pScratch1 += 4u;
emilmont 1:fdd22bb7aa52 328
emilmont 1:fdd22bb7aa52 329 }
emilmont 1:fdd22bb7aa52 330
emilmont 1:fdd22bb7aa52 331
emilmont 1:fdd22bb7aa52 332 blkCnt = numPoints & 0x3;
emilmont 1:fdd22bb7aa52 333
emilmont 1:fdd22bb7aa52 334 /* Calculate convolution for remaining samples of Bigger length sequence */
emilmont 1:fdd22bb7aa52 335 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 336 {
emilmont 1:fdd22bb7aa52 337 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 338 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 339
emilmont 1:fdd22bb7aa52 340 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 341 acc0 = 0;
emilmont 1:fdd22bb7aa52 342
emilmont 1:fdd22bb7aa52 343 tapCnt = (srcBLen) >> 1u;
emilmont 1:fdd22bb7aa52 344
emilmont 1:fdd22bb7aa52 345 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 346 {
emilmont 1:fdd22bb7aa52 347
emilmont 1:fdd22bb7aa52 348 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 349 x1 = *__SIMD32(pScr1)++;
emilmont 1:fdd22bb7aa52 350
emilmont 1:fdd22bb7aa52 351 /* Read two samples from smaller buffer */
emilmont 1:fdd22bb7aa52 352 y1 = *__SIMD32(pIn2)++;
emilmont 1:fdd22bb7aa52 353
emilmont 1:fdd22bb7aa52 354 acc0 = __SMLALD(x1, y1, acc0);
emilmont 1:fdd22bb7aa52 355
emilmont 1:fdd22bb7aa52 356 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 357 tapCnt--;
emilmont 1:fdd22bb7aa52 358 }
emilmont 1:fdd22bb7aa52 359
emilmont 1:fdd22bb7aa52 360 tapCnt = (srcBLen) & 1u;
emilmont 1:fdd22bb7aa52 361
emilmont 1:fdd22bb7aa52 362 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 363 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 364 {
emilmont 1:fdd22bb7aa52 365
emilmont 1:fdd22bb7aa52 366 /* accumlate the results */
emilmont 1:fdd22bb7aa52 367 acc0 += (*pScr1++ * *pIn2++);
emilmont 1:fdd22bb7aa52 368
emilmont 1:fdd22bb7aa52 369 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 370 tapCnt--;
emilmont 1:fdd22bb7aa52 371 }
emilmont 1:fdd22bb7aa52 372
emilmont 1:fdd22bb7aa52 373 blkCnt--;
emilmont 1:fdd22bb7aa52 374
emilmont 1:fdd22bb7aa52 375 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 376 *pOut++ = (q15_t) (__SSAT((acc0 >> 15), 16));
emilmont 1:fdd22bb7aa52 377
emilmont 1:fdd22bb7aa52 378 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 379 pIn2 = py;
emilmont 1:fdd22bb7aa52 380
emilmont 1:fdd22bb7aa52 381 pScratch1 += 1u;
emilmont 1:fdd22bb7aa52 382
emilmont 1:fdd22bb7aa52 383 }
emilmont 1:fdd22bb7aa52 384
emilmont 1:fdd22bb7aa52 385 /* set status as ARM_MATH_SUCCESS */
emilmont 1:fdd22bb7aa52 386 status = ARM_MATH_SUCCESS;
emilmont 1:fdd22bb7aa52 387
emilmont 1:fdd22bb7aa52 388 }
emilmont 1:fdd22bb7aa52 389
emilmont 1:fdd22bb7aa52 390 /* Return to application */
emilmont 1:fdd22bb7aa52 391 return (status);
emilmont 1:fdd22bb7aa52 392 }
emilmont 1:fdd22bb7aa52 393
emilmont 1:fdd22bb7aa52 394 #else
emilmont 1:fdd22bb7aa52 395
emilmont 1:fdd22bb7aa52 396 arm_status arm_conv_partial_opt_q15(
emilmont 1:fdd22bb7aa52 397 q15_t * pSrcA,
emilmont 1:fdd22bb7aa52 398 uint32_t srcALen,
emilmont 1:fdd22bb7aa52 399 q15_t * pSrcB,
emilmont 1:fdd22bb7aa52 400 uint32_t srcBLen,
emilmont 1:fdd22bb7aa52 401 q15_t * pDst,
emilmont 1:fdd22bb7aa52 402 uint32_t firstIndex,
emilmont 1:fdd22bb7aa52 403 uint32_t numPoints,
emilmont 1:fdd22bb7aa52 404 q15_t * pScratch1,
emilmont 1:fdd22bb7aa52 405 q15_t * pScratch2)
emilmont 1:fdd22bb7aa52 406 {
emilmont 1:fdd22bb7aa52 407
emilmont 1:fdd22bb7aa52 408 q15_t *pOut = pDst; /* output pointer */
emilmont 1:fdd22bb7aa52 409 q15_t *pScr1 = pScratch1; /* Temporary pointer for scratch1 */
emilmont 1:fdd22bb7aa52 410 q15_t *pScr2 = pScratch2; /* Temporary pointer for scratch1 */
emilmont 1:fdd22bb7aa52 411 q63_t acc0, acc1, acc2, acc3; /* Accumulator */
emilmont 1:fdd22bb7aa52 412 q15_t *pIn1; /* inputA pointer */
emilmont 1:fdd22bb7aa52 413 q15_t *pIn2; /* inputB pointer */
emilmont 1:fdd22bb7aa52 414 q15_t *px; /* Intermediate inputA pointer */
emilmont 1:fdd22bb7aa52 415 q15_t *py; /* Intermediate inputB pointer */
emilmont 1:fdd22bb7aa52 416 uint32_t j, k, blkCnt; /* loop counter */
emilmont 1:fdd22bb7aa52 417 arm_status status; /* Status variable */
emilmont 1:fdd22bb7aa52 418 uint32_t tapCnt; /* loop count */
emilmont 1:fdd22bb7aa52 419 q15_t x10, x11, x20, x21; /* Temporary variables to hold srcA buffer */
emilmont 1:fdd22bb7aa52 420 q15_t y10, y11; /* Temporary variables to hold srcB buffer */
emilmont 1:fdd22bb7aa52 421
emilmont 1:fdd22bb7aa52 422
emilmont 1:fdd22bb7aa52 423 /* Check for range of output samples to be calculated */
emilmont 1:fdd22bb7aa52 424 if((firstIndex + numPoints) > ((srcALen + (srcBLen - 1u))))
emilmont 1:fdd22bb7aa52 425 {
emilmont 1:fdd22bb7aa52 426 /* Set status as ARM_MATH_ARGUMENT_ERROR */
emilmont 1:fdd22bb7aa52 427 status = ARM_MATH_ARGUMENT_ERROR;
emilmont 1:fdd22bb7aa52 428 }
emilmont 1:fdd22bb7aa52 429 else
emilmont 1:fdd22bb7aa52 430 {
emilmont 1:fdd22bb7aa52 431
emilmont 1:fdd22bb7aa52 432 /* The algorithm implementation is based on the lengths of the inputs. */
emilmont 1:fdd22bb7aa52 433 /* srcB is always made to slide across srcA. */
emilmont 1:fdd22bb7aa52 434 /* So srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 435 if(srcALen >= srcBLen)
emilmont 1:fdd22bb7aa52 436 {
emilmont 1:fdd22bb7aa52 437 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 438 pIn1 = pSrcA;
emilmont 1:fdd22bb7aa52 439
emilmont 1:fdd22bb7aa52 440 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 441 pIn2 = pSrcB;
emilmont 1:fdd22bb7aa52 442 }
emilmont 1:fdd22bb7aa52 443 else
emilmont 1:fdd22bb7aa52 444 {
emilmont 1:fdd22bb7aa52 445 /* Initialization of inputA pointer */
emilmont 1:fdd22bb7aa52 446 pIn1 = pSrcB;
emilmont 1:fdd22bb7aa52 447
emilmont 1:fdd22bb7aa52 448 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 449 pIn2 = pSrcA;
emilmont 1:fdd22bb7aa52 450
emilmont 1:fdd22bb7aa52 451 /* srcBLen is always considered as shorter or equal to srcALen */
emilmont 1:fdd22bb7aa52 452 j = srcBLen;
emilmont 1:fdd22bb7aa52 453 srcBLen = srcALen;
emilmont 1:fdd22bb7aa52 454 srcALen = j;
emilmont 1:fdd22bb7aa52 455 }
emilmont 1:fdd22bb7aa52 456
emilmont 1:fdd22bb7aa52 457 /* Temporary pointer for scratch2 */
emilmont 1:fdd22bb7aa52 458 py = pScratch2;
emilmont 1:fdd22bb7aa52 459
emilmont 1:fdd22bb7aa52 460 /* pointer to take end of scratch2 buffer */
emilmont 1:fdd22bb7aa52 461 pScr2 = pScratch2 + srcBLen - 1;
emilmont 1:fdd22bb7aa52 462
emilmont 1:fdd22bb7aa52 463 /* points to smaller length sequence */
emilmont 1:fdd22bb7aa52 464 px = pIn2;
emilmont 1:fdd22bb7aa52 465
emilmont 1:fdd22bb7aa52 466 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 467 k = srcBLen >> 2u;
emilmont 1:fdd22bb7aa52 468
emilmont 1:fdd22bb7aa52 469 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 470 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 471 while(k > 0u)
emilmont 1:fdd22bb7aa52 472 {
emilmont 1:fdd22bb7aa52 473 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 474 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 475 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 476 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 477 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 478
emilmont 1:fdd22bb7aa52 479 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 480 k--;
emilmont 1:fdd22bb7aa52 481 }
emilmont 1:fdd22bb7aa52 482
emilmont 1:fdd22bb7aa52 483 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 484 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 485 k = srcBLen % 0x4u;
emilmont 1:fdd22bb7aa52 486
emilmont 1:fdd22bb7aa52 487 while(k > 0u)
emilmont 1:fdd22bb7aa52 488 {
emilmont 1:fdd22bb7aa52 489 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 490 *pScr2-- = *px++;
emilmont 1:fdd22bb7aa52 491
emilmont 1:fdd22bb7aa52 492 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 493 k--;
emilmont 1:fdd22bb7aa52 494 }
emilmont 1:fdd22bb7aa52 495
emilmont 1:fdd22bb7aa52 496 /* Initialze temporary scratch pointer */
emilmont 1:fdd22bb7aa52 497 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 498
emilmont 1:fdd22bb7aa52 499 /* Fill (srcBLen - 1u) zeros in scratch buffer */
emilmont 1:fdd22bb7aa52 500 arm_fill_q15(0, pScr1, (srcBLen - 1u));
emilmont 1:fdd22bb7aa52 501
emilmont 1:fdd22bb7aa52 502 /* Update temporary scratch pointer */
emilmont 1:fdd22bb7aa52 503 pScr1 += (srcBLen - 1u);
emilmont 1:fdd22bb7aa52 504
emilmont 1:fdd22bb7aa52 505 /* Copy bigger length sequence(srcALen) samples in scratch1 buffer */
emilmont 1:fdd22bb7aa52 506
emilmont 1:fdd22bb7aa52 507
emilmont 1:fdd22bb7aa52 508 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 509 k = srcALen >> 2u;
emilmont 1:fdd22bb7aa52 510
emilmont 1:fdd22bb7aa52 511 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 512 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 513 while(k > 0u)
emilmont 1:fdd22bb7aa52 514 {
emilmont 1:fdd22bb7aa52 515 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 516 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 517 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 518 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 519 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 520
emilmont 1:fdd22bb7aa52 521 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 522 k--;
emilmont 1:fdd22bb7aa52 523 }
emilmont 1:fdd22bb7aa52 524
emilmont 1:fdd22bb7aa52 525 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 526 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 527 k = srcALen % 0x4u;
emilmont 1:fdd22bb7aa52 528
emilmont 1:fdd22bb7aa52 529 while(k > 0u)
emilmont 1:fdd22bb7aa52 530 {
emilmont 1:fdd22bb7aa52 531 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 532 *pScr1++ = *pIn1++;
emilmont 1:fdd22bb7aa52 533
emilmont 1:fdd22bb7aa52 534 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 535 k--;
emilmont 1:fdd22bb7aa52 536 }
emilmont 1:fdd22bb7aa52 537
emilmont 1:fdd22bb7aa52 538
emilmont 1:fdd22bb7aa52 539 /* Apply loop unrolling and do 4 Copies simultaneously. */
emilmont 1:fdd22bb7aa52 540 k = (srcBLen - 1u) >> 2u;
emilmont 1:fdd22bb7aa52 541
emilmont 1:fdd22bb7aa52 542 /* First part of the processing with loop unrolling copies 4 data points at a time.
emilmont 1:fdd22bb7aa52 543 ** a second loop below copies for the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 544 while(k > 0u)
emilmont 1:fdd22bb7aa52 545 {
emilmont 1:fdd22bb7aa52 546 /* copy second buffer in reversal manner */
emilmont 1:fdd22bb7aa52 547 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 548 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 549 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 550 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 551
emilmont 1:fdd22bb7aa52 552 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 553 k--;
emilmont 1:fdd22bb7aa52 554 }
emilmont 1:fdd22bb7aa52 555
emilmont 1:fdd22bb7aa52 556 /* If the count is not a multiple of 4, copy remaining samples here.
emilmont 1:fdd22bb7aa52 557 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 558 k = (srcBLen - 1u) % 0x4u;
emilmont 1:fdd22bb7aa52 559
emilmont 1:fdd22bb7aa52 560 while(k > 0u)
emilmont 1:fdd22bb7aa52 561 {
emilmont 1:fdd22bb7aa52 562 /* copy second buffer in reversal manner for remaining samples */
emilmont 1:fdd22bb7aa52 563 *pScr1++ = 0;
emilmont 1:fdd22bb7aa52 564
emilmont 1:fdd22bb7aa52 565 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 566 k--;
emilmont 1:fdd22bb7aa52 567 }
emilmont 1:fdd22bb7aa52 568
emilmont 1:fdd22bb7aa52 569
emilmont 1:fdd22bb7aa52 570 /* Initialization of pIn2 pointer */
emilmont 1:fdd22bb7aa52 571 pIn2 = py;
emilmont 1:fdd22bb7aa52 572
emilmont 1:fdd22bb7aa52 573 pScratch1 += firstIndex;
emilmont 1:fdd22bb7aa52 574
emilmont 1:fdd22bb7aa52 575 pOut = pDst + firstIndex;
emilmont 1:fdd22bb7aa52 576
emilmont 1:fdd22bb7aa52 577 /* Actual convolution process starts here */
emilmont 1:fdd22bb7aa52 578 blkCnt = (numPoints) >> 2;
emilmont 1:fdd22bb7aa52 579
emilmont 1:fdd22bb7aa52 580 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 581 {
emilmont 1:fdd22bb7aa52 582 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 583 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 584
emilmont 1:fdd22bb7aa52 585 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 586 acc0 = 0;
emilmont 1:fdd22bb7aa52 587 acc1 = 0;
emilmont 1:fdd22bb7aa52 588 acc2 = 0;
emilmont 1:fdd22bb7aa52 589 acc3 = 0;
emilmont 1:fdd22bb7aa52 590
emilmont 1:fdd22bb7aa52 591 /* Read two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 592 x10 = *pScr1++;
emilmont 1:fdd22bb7aa52 593 x11 = *pScr1++;
emilmont 1:fdd22bb7aa52 594
emilmont 1:fdd22bb7aa52 595 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 596 x20 = *pScr1++;
emilmont 1:fdd22bb7aa52 597 x21 = *pScr1++;
emilmont 1:fdd22bb7aa52 598
emilmont 1:fdd22bb7aa52 599 tapCnt = (srcBLen) >> 2u;
emilmont 1:fdd22bb7aa52 600
emilmont 1:fdd22bb7aa52 601 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 602 {
emilmont 1:fdd22bb7aa52 603
emilmont 1:fdd22bb7aa52 604 /* Read two samples from smaller buffer */
emilmont 1:fdd22bb7aa52 605 y10 = *pIn2;
emilmont 1:fdd22bb7aa52 606 y11 = *(pIn2 + 1u);
emilmont 1:fdd22bb7aa52 607
emilmont 1:fdd22bb7aa52 608 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 609 acc0 += (q63_t) x10 *y10;
emilmont 1:fdd22bb7aa52 610 acc0 += (q63_t) x11 *y11;
emilmont 1:fdd22bb7aa52 611 acc2 += (q63_t) x20 *y10;
emilmont 1:fdd22bb7aa52 612 acc2 += (q63_t) x21 *y11;
emilmont 1:fdd22bb7aa52 613
emilmont 1:fdd22bb7aa52 614 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 615 acc1 += (q63_t) x11 *y10;
emilmont 1:fdd22bb7aa52 616 acc1 += (q63_t) x20 *y11;
emilmont 1:fdd22bb7aa52 617
emilmont 1:fdd22bb7aa52 618 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 619 x10 = *pScr1;
emilmont 1:fdd22bb7aa52 620 x11 = *(pScr1 + 1u);
emilmont 1:fdd22bb7aa52 621
emilmont 1:fdd22bb7aa52 622 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 623 acc3 += (q63_t) x21 *y10;
emilmont 1:fdd22bb7aa52 624 acc3 += (q63_t) x10 *y11;
emilmont 1:fdd22bb7aa52 625
emilmont 1:fdd22bb7aa52 626 /* Read next two samples from scratch2 buffer */
emilmont 1:fdd22bb7aa52 627 y10 = *(pIn2 + 2u);
emilmont 1:fdd22bb7aa52 628 y11 = *(pIn2 + 3u);
emilmont 1:fdd22bb7aa52 629
emilmont 1:fdd22bb7aa52 630 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 631 acc0 += (q63_t) x20 *y10;
emilmont 1:fdd22bb7aa52 632 acc0 += (q63_t) x21 *y11;
emilmont 1:fdd22bb7aa52 633 acc2 += (q63_t) x10 *y10;
emilmont 1:fdd22bb7aa52 634 acc2 += (q63_t) x11 *y11;
emilmont 1:fdd22bb7aa52 635 acc1 += (q63_t) x21 *y10;
emilmont 1:fdd22bb7aa52 636 acc1 += (q63_t) x10 *y11;
emilmont 1:fdd22bb7aa52 637
emilmont 1:fdd22bb7aa52 638 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 639 x20 = *(pScr1 + 2);
emilmont 1:fdd22bb7aa52 640 x21 = *(pScr1 + 3);
emilmont 1:fdd22bb7aa52 641
emilmont 1:fdd22bb7aa52 642 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 643 acc3 += (q63_t) x11 *y10;
emilmont 1:fdd22bb7aa52 644 acc3 += (q63_t) x20 *y11;
emilmont 1:fdd22bb7aa52 645
emilmont 1:fdd22bb7aa52 646 /* update scratch pointers */
emilmont 1:fdd22bb7aa52 647 pIn2 += 4u;
emilmont 1:fdd22bb7aa52 648 pScr1 += 4u;
emilmont 1:fdd22bb7aa52 649
emilmont 1:fdd22bb7aa52 650 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 651 tapCnt--;
emilmont 1:fdd22bb7aa52 652 }
emilmont 1:fdd22bb7aa52 653
emilmont 1:fdd22bb7aa52 654 /* Update scratch pointer for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 655 pScr1 -= 4u;
emilmont 1:fdd22bb7aa52 656
emilmont 1:fdd22bb7aa52 657 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 658 tapCnt = (srcBLen) & 3u;
emilmont 1:fdd22bb7aa52 659
emilmont 1:fdd22bb7aa52 660 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 661 {
emilmont 1:fdd22bb7aa52 662 /* accumlate the results */
emilmont 1:fdd22bb7aa52 663 acc0 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 664 acc1 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 665 acc2 += (*pScr1++ * *pIn2);
emilmont 1:fdd22bb7aa52 666 acc3 += (*pScr1++ * *pIn2++);
emilmont 1:fdd22bb7aa52 667
emilmont 1:fdd22bb7aa52 668 pScr1 -= 3u;
emilmont 1:fdd22bb7aa52 669
emilmont 1:fdd22bb7aa52 670 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 671 tapCnt--;
emilmont 1:fdd22bb7aa52 672 }
emilmont 1:fdd22bb7aa52 673
emilmont 1:fdd22bb7aa52 674 blkCnt--;
emilmont 1:fdd22bb7aa52 675
emilmont 1:fdd22bb7aa52 676
emilmont 1:fdd22bb7aa52 677 /* Store the results in the accumulators in the destination buffer. */
emilmont 1:fdd22bb7aa52 678 *pOut++ = __SSAT((acc0 >> 15), 16);
emilmont 1:fdd22bb7aa52 679 *pOut++ = __SSAT((acc1 >> 15), 16);
emilmont 1:fdd22bb7aa52 680 *pOut++ = __SSAT((acc2 >> 15), 16);
emilmont 1:fdd22bb7aa52 681 *pOut++ = __SSAT((acc3 >> 15), 16);
emilmont 1:fdd22bb7aa52 682
emilmont 1:fdd22bb7aa52 683
emilmont 1:fdd22bb7aa52 684 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 685 pIn2 = py;
emilmont 1:fdd22bb7aa52 686
emilmont 1:fdd22bb7aa52 687 pScratch1 += 4u;
emilmont 1:fdd22bb7aa52 688
emilmont 1:fdd22bb7aa52 689 }
emilmont 1:fdd22bb7aa52 690
emilmont 1:fdd22bb7aa52 691
emilmont 1:fdd22bb7aa52 692 blkCnt = numPoints & 0x3;
emilmont 1:fdd22bb7aa52 693
emilmont 1:fdd22bb7aa52 694 /* Calculate convolution for remaining samples of Bigger length sequence */
emilmont 1:fdd22bb7aa52 695 while(blkCnt > 0)
emilmont 1:fdd22bb7aa52 696 {
emilmont 1:fdd22bb7aa52 697 /* Initialze temporary scratch pointer as scratch1 */
emilmont 1:fdd22bb7aa52 698 pScr1 = pScratch1;
emilmont 1:fdd22bb7aa52 699
emilmont 1:fdd22bb7aa52 700 /* Clear Accumlators */
emilmont 1:fdd22bb7aa52 701 acc0 = 0;
emilmont 1:fdd22bb7aa52 702
emilmont 1:fdd22bb7aa52 703 tapCnt = (srcBLen) >> 1u;
emilmont 1:fdd22bb7aa52 704
emilmont 1:fdd22bb7aa52 705 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 706 {
emilmont 1:fdd22bb7aa52 707
emilmont 1:fdd22bb7aa52 708 /* Read next two samples from scratch1 buffer */
emilmont 1:fdd22bb7aa52 709 x10 = *pScr1++;
emilmont 1:fdd22bb7aa52 710 x11 = *pScr1++;
emilmont 1:fdd22bb7aa52 711
emilmont 1:fdd22bb7aa52 712 /* Read two samples from smaller buffer */
emilmont 1:fdd22bb7aa52 713 y10 = *pIn2++;
emilmont 1:fdd22bb7aa52 714 y11 = *pIn2++;
emilmont 1:fdd22bb7aa52 715
emilmont 1:fdd22bb7aa52 716 /* multiply and accumlate */
emilmont 1:fdd22bb7aa52 717 acc0 += (q63_t) x10 *y10;
emilmont 1:fdd22bb7aa52 718 acc0 += (q63_t) x11 *y11;
emilmont 1:fdd22bb7aa52 719
emilmont 1:fdd22bb7aa52 720 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 721 tapCnt--;
emilmont 1:fdd22bb7aa52 722 }
emilmont 1:fdd22bb7aa52 723
emilmont 1:fdd22bb7aa52 724 tapCnt = (srcBLen) & 1u;
emilmont 1:fdd22bb7aa52 725
emilmont 1:fdd22bb7aa52 726 /* apply same above for remaining samples of smaller length sequence */
emilmont 1:fdd22bb7aa52 727 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 728 {
emilmont 1:fdd22bb7aa52 729
emilmont 1:fdd22bb7aa52 730 /* accumlate the results */
emilmont 1:fdd22bb7aa52 731 acc0 += (*pScr1++ * *pIn2++);
emilmont 1:fdd22bb7aa52 732
emilmont 1:fdd22bb7aa52 733 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 734 tapCnt--;
emilmont 1:fdd22bb7aa52 735 }
emilmont 1:fdd22bb7aa52 736
emilmont 1:fdd22bb7aa52 737 blkCnt--;
emilmont 1:fdd22bb7aa52 738
emilmont 1:fdd22bb7aa52 739 /* Store the result in the accumulator in the destination buffer. */
emilmont 1:fdd22bb7aa52 740 *pOut++ = (q15_t) (__SSAT((acc0 >> 15), 16));
emilmont 1:fdd22bb7aa52 741
emilmont 1:fdd22bb7aa52 742
emilmont 1:fdd22bb7aa52 743 /* Initialization of inputB pointer */
emilmont 1:fdd22bb7aa52 744 pIn2 = py;
emilmont 1:fdd22bb7aa52 745
emilmont 1:fdd22bb7aa52 746 pScratch1 += 1u;
emilmont 1:fdd22bb7aa52 747
emilmont 1:fdd22bb7aa52 748 }
emilmont 1:fdd22bb7aa52 749
emilmont 1:fdd22bb7aa52 750 /* set status as ARM_MATH_SUCCESS */
emilmont 1:fdd22bb7aa52 751 status = ARM_MATH_SUCCESS;
emilmont 1:fdd22bb7aa52 752
emilmont 1:fdd22bb7aa52 753 }
emilmont 1:fdd22bb7aa52 754
emilmont 1:fdd22bb7aa52 755 /* Return to application */
emilmont 1:fdd22bb7aa52 756 return (status);
emilmont 1:fdd22bb7aa52 757 }
emilmont 1:fdd22bb7aa52 758
emilmont 1:fdd22bb7aa52 759 #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */
emilmont 1:fdd22bb7aa52 760
emilmont 1:fdd22bb7aa52 761
emilmont 1:fdd22bb7aa52 762 /**
emilmont 1:fdd22bb7aa52 763 * @} end of PartialConv group
emilmont 1:fdd22bb7aa52 764 */